A real-time two-stage and dual-check template matching algorithm based on normalized cross-correlation for industrial vision positioning

In this paper, a fast template matching algorithm of two-stage and dual-check bounded partial correlation (TDBPC) based on normalized cross-correlation (NCC) of single-check bounded partial correlation is proposed. According to the principle of continuous rows, the template and the sub-image under matching are divided into three subregions to obtain two upper boundary terms of NCC and get two checking conditions then. In this way, it is possible to quickly eliminate matching points that cannot provide a better cross-correlation score regarding the current best candidate. Generally, to get the highest cross-correlation score, the sub-image has to traverse through the whole image. In addition, the two-stage search strategy of coarse–fine proposed in this paper can further reduce the calculation and improve matching efficiency. The initialization parameters are selected experimentally or automatically. Experimental results show that the TDBPC algorithm proposed in this paper can solve high computational complexity and long matching time of NCC template matching and make it possible to achieve real-time template matching in industrial vision positioning fields. The feasibility of this algorithm in practical application is proved.